LCR riaa's

the road not taken.
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

LCR riaa's

Post by dave slagle »

i finally got around to plugging the tango (pultec) into spice and started playing to see what the issues were on paper with scaling them up. I was quickly hit in the forehead with my palm when i realized the whole issue with the 600R units was the load it provided to what was driving them and not the units themself.

the design requires either a 600R source or a 600R load to keep things happy. I'll state now that the terminaring resistor is manditory and trying to use a source=load situation is futile.

looking at the choke values needed for the 6K unit, i set out to do them and see where things go. it quickly becomes apparent in spice that small value changes in L with frequency can really screw things up, but armed with the experience of winding sB's Lriaa chokes, it only took a few revisions to get something that should behave in circuit as it does in spice.

as impedance of the LCR goes up so does variability of the response so the choice of tradeoff is an important one.

at some point i should be able to get a LT spice file and an excel file up that will figure all of the values out and plot the response (with an iRiaa) and i think the design methodology should be pretty simple.

Pick your tube
Pick your load (multiple of Rp at operating point)
Pick your LCR impedance to match your desired load.
Understand that a 27 into a 20K LCR is not a wise choice.

dave
DowdyLama
Posts: 104
Joined: Thu Jan 27, 2005 5:42 pm
Location: Blue Ridge Mtns, NC
Contact:

Post by DowdyLama »

Assuming that we're sticking with the 600 ohm units for now [with a 600 R termination]:
how does it look with source = 1/2 load? or even
source = 1/3 load?
IslandPink
Posts: 56
Joined: Thu May 12, 2005 8:01 am
Location: Denbigh, North Wales

Pentode

Post by IslandPink »

Dave

Related question - does Thorsten's solution of driving it with a pentode ( E810F ) seem like a logical solution - can you model it ? - seems to work in practice from what I heard of Nick Gorham's phono unit in the UK .

Mark
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

DowdyLama wrote:Assuming that we're sticking with the 600 ohm units for now [with a 600 R termination]:
how does it look with source = 1/2 load? or even
source = 1/3 load?
thinking of the 421A are we :-)

from what i could tell the source impedance didn't have much effect on the shape of the filter, but it did on the amlitude.

If we accept that the Rp of a class A tube varies dynamically, we also have to accept that the source impedance varied dynamically.

The lower the ratio of source to load is the more this effect will show up. I'm not sure it would be easy to measure or see this effect, but conceptually it could explain a lot. If you consider the LCR a precision filter, (people talk about .1dB accuracy) anything that reduces variations should be a good thing. This doesn't mean that you need to drive a 600R lcr with 6 ohms, but i would consider the ratios chosen in the other parts of the amplification chain and go from there.

dave
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Re: Pentode

Post by dave slagle »

IslandPink wrote:Dave

Related question - does Thorsten's solution of driving it with a pentode ( E810F ) seem like a logical solution - can you model it ? - seems to work in practice from what I heard of Nick Gorham's phono unit in the UK .

Mark
not sure if i have the spice model for the E810F i'll look. On paper the pentode seems like a brilliant answer for this situation since the load mostly resistive. The sound of the few pentode driver circuits i have built/heard have not inspired me to follow that path. (of course the E810F could be triode wired and the same load should work out well for either and you can make up your own mind)

dave
DowdyLama
Posts: 104
Joined: Thu Jan 27, 2005 5:42 pm
Location: Blue Ridge Mtns, NC
Contact:

Post by DowdyLama »

"thinking of the 421A are we?"

I'm certainly thinking of using either the 421A and/or the 6AS7 into a 1:1 here - my extremely generalized experience [whether considering the output z of a linestage or an A2 driver] has been that we often pursue (the theoretical notion) of low impedance at the expense of sonics/common sense.
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

from a source to load viewpoint that seems to be the only tube that will work. the lack of gain troubles me a bit, which is why i looked towards a higher mu tube.

I do think i would take a AS7 through a 1:1 into 600R over a CF, but a 5687 through an 8:1 sure would be appealing on paper and give you the same net gain (loss). The tradeoff becomes the tube sonics and the difference between an 8:1 with a l10+X source ot load ratio and a 1:1 with a 3X ratio. I'll call it a coin toss, but the 1:1 is much more appealing from the iron viewpoint.

The 6C45 right into a 6K unit will actually give you some gain and an AV20 into a 6K will have the same source to load ratio as your AS7 @ 600R but 10X the gain.


dave
IslandPink
Posts: 56
Joined: Thu May 12, 2005 8:01 am
Location: Denbigh, North Wales

Pentode

Post by IslandPink »

For info on the pentode option he had two circuits. ( I can't access the links here at work but can paste later if I get chance ) . The 2002 circuit used E810F with about 8.2K load driving the input, and terminated with 600R at the far end . The 2005 circuit again used E810 but with a very low ( 600R + a bit ) load on the pentode driver and no significant load on the output . I heard the former option and it was clearly very good , but haven't made detailed comparisons with anything similar yet . This was using the S&B 600R units .

I'm soon to try my D3a /6GK5 phono with the triode D3a converted into Pentode to drive the ( normal RC )RIAA , so I'll comment on how that sounds if it's interesting .

MJ
DowdyLama
Posts: 104
Joined: Thu Jan 27, 2005 5:42 pm
Location: Blue Ridge Mtns, NC
Contact:

Post by DowdyLama »

"The tradeoff becomes the tube sonics and the difference between an 8:1 with a 10+X source ot load ratio and a 1:1 with a 3X ratio...but the 1:1 is much more appealing from the iron viewpoint."

Absolutely - and that's my whole point!
I'm going to continue to consider the monkeyed 6C45-6AS7 with a Slagle 1:1...
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Dave and Mark,

Apologies if I'm intruding.

I think Dave is spot on, the problem is not the 600R EQ unit per se but driving it.

Thorsten's E810F is one solution for a 600R driver, and I've heard NickG's unit as well. However, he currently uses a 600R terminating load; so his design is more like Thorsten's first version and is as per Dave's suggestion.

I've been playing with two alternatives.

First off is a 6C45PE Cascode.

The other is cathode follower based.

I like the idea of something in between the 600R and the 10k offered by S&B. The challenge with the 600R is, as you say, its very hard to drive. The other problem S&B have found, I believe is ensuring the 30H choke required by the 10K unit is flat across the full frequency range. I think they will admit, when pushed, that the EQ on 10K version is not as good as their 600R. A compromise in betwixt might be ideal as it would open up a larger range of valves without the cost/challenges of a 30H choke.

I'm planning tryinga all three designs E180F, cascode and CF, but need to find the cash for the cans first, until then, my experiments are purely based in Spice.

Unfortunately, I don't have a spice model for the E810F, but if your using it in Pentode mode for a 600R source design, I think it could be modeled with voltage source with a series resistance of 600R.

cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Oh, its probably worth adding that the the cathode follower version could, in theory, support a 200R matched EQ, now whether this is a good idea, well......?

cheers,

-- Andrew
James D
Posts: 15
Joined: Thu Mar 31, 2005 2:00 pm
Location: UK

Post by James D »

I would have thought that an Aikido front end was the ideal way to drive the 600R Eq... A pentode Aikido would give you an easy 100-150 times gain and an ECC83 front end aikido would give you 50 times.

The output impedance is nice and stable and as a series resistance would be needed to get the 600R Zout the aging variation of the Zout would be small...

Just a thought. Andrew and I have modelled this a bit and conceptually quite like the Cascode with 6C45s as the GM is high and Va can be kept lower so minimising noise... but the Aikido is probably more linear...

James
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Yes, James is spot on the Cathode Follower version is based on JB's Aikido. I should have said. The LT Spice model I have is for a 6072A -> 6N30-> EQ600 -> D3A triode connected. I like the 6072A! I loaded the D3A with a constant current source and get a nice 40db gain. Since this is an Aikido the tubes can be swapped easily an ECC83 version would give ~49db.

The 6C45 cascode was all James's fault, I just did the modelling.

I have yet to run a sim for the 6C45 into EQ-600, the model is just the front end. I will do this when I have time.

The Aikdo appears very linear indeed.

The big issue with these two circuits is the >3 amp DC heater requirements, which I have now solved using a 5 amp LM1085.

I was thinking of perhaps trying a Pentode for the second stage where noise is less of an issue. Why? Well Iwould like to try this and see how it compares to triode back-end with stepups for the extra gain.

Saving my pennies for the cans at the moment.

Will post finding on the 6C45 when I have a completed model.

cheers,

-- Andrew
nickg
Posts: 29
Joined: Sat Mar 19, 2005 1:46 pm

Post by nickg »

Argh, I have been looking in on the forum on a regular basis, and thinking that there were no posts and it was all a bit quiet. I just noticed my browser had decided to stop logging me in, so I had missed all these posts about RIAA. Poo.

Thats it, I will go away again now :-)
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Argh, I have been looking in on the forum on a regular basis, and thinking that there were no posts and it was all a bit quiet. I just noticed my browser had decided to stop logging me in, so I had missed all these posts about RIAA. Poo.

Thats it, I will go away again now :-)
Oh Nick don't go away......

I was also hoping Dave might chip in....Dave any thoughts/comments?

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I have been meaning to chime in, but i haven't had much to say. On paper i cannot argue with the approach and to be honest i haven't played enough with pentodes to give an informed answer. The little bit i have done with pentodes has not encouraged me to follow down that road. Of course that doesn't mean given more effort good results can not be had, just that i was lazy and chose the easy way out. :-)

dave
nickg
Posts: 29
Joined: Sat Mar 19, 2005 1:46 pm

Post by nickg »

I am not at work next week, and was planning to get set up for some loudspeaker measuring. I should be able to do some tests on the pentode driver LCR phono I am using, to try and get an idea what the S/N ratio is. At the moment using a Denon 103 (0.3mv) and 1:20 TX's noise is certainly not an audible problem.

Would a FFT of its output, driving it with a tone from a test disk provide any useful information?
Last edited by nickg on Thu Aug 10, 2006 8:59 pm, edited 1 time in total.
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Dave et al,

OK I'll pipe up with a few specifics that I think might tempt you.

1) What's your opinion on the 6C45 cascode? I think this will be a very worthy alternative to the E810F.

2) Any thoughts on the Aikido and specifically what about a 200R RIAA EQ? The Cathode Followers could drive this matched impedance, of course, that's not to say doing so would be a good idea. Just seeking opinions.

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Hey,

I have no experience wiht cascodes so i'll defer to the others. I also have not played or even looked into the akido, but have to ask if the Z-out is stable enough to operate in a matched situation.

WRT the 200 ohm LCR, I haven't built that either (what good am i?) but it is interesting to note the tradeoffs in part selection. For the 200 ohm unit the inductors become really easy to get done (600mhy and 15mhy) but the caps get big (15uf and .4uf). One of the perceived benefits in going to 6K (assuming the inductors don't bite you) is the caps become quite small (.5uf and 12.6nf) and if we subscribe to the smaller caps sound better this might be heading down the right path. So maybe our goal is ot find the ultimate compromise between inductor and cap selection and then make that impedance work.

oddly enough both the 600 ohm and the 10K versions have offbeat cap sizes. Sure for the 600 ohm a 5 mic cap can get you 5.02 but the .126 requires some work. Going to 10K requires a 3.01 (again easy) and a 7.56nf (again a problem). Why not choose a rational number like 7k56 for the impedance? This suddenly makes your cap choices .4uf and 10nf both easy values to come up in a multitude of types.

I guess what i need is some crazed cap-o-phile (with a good cap meter) to roll caps through a design and verify if there are major sonic signatures in that particular position.

dave
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Dave,

All good points. I never for a moment wondered why 600R was historicallychosen, commercial standards and all that, but 10K, with all the issues around the larger inductor, seemed an odd choice.

The Aikido, Pentode and Casode solutions can all be (re)designed to match any impendance, within reason. In fact the the Pentode and Cascode solutions should work just as well, if not better with slighly higher anode load. The Aikido solution is impedance matched with a series resistor; this is can be both good and bad but does make the RIAA almost immune to valve ageing.

There seems to be some credence given to the possibility that lower impedance RIAA networks might offer better sound, assuming the circuit can be driven correctly. And, as a paper exercise, it might be interesting to try slighly higher values that produce 'sensible' cap sizes. I'm just offering this a a counter balance to the lower cap sounds better camp; tho' I am a subscriber to that point of view.

As you say, what is the best balance? How much higher than 600R is better for sound is the $60,000,000 question? Certainly at 6K you avoid many of the problems that the Pentode, Aikido and Cascode circuits are designed to solve.

Out of interest, would you be willing to publish the ratios for the LCR RIAA your using? They seem slightly different to mine, which were gleemed form some circuits I found on the web, but might not be completely accurate.

I'd also be intersted in hearing your choice of valve you had in mind for 6K?

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Andrew wrote: There seems to be some credence given to the possibility that lower impedance RIAA networks might offer better sound,
what is your theory behind this? just curious...
Out of interest, would you be willing to publish the ratios for the LCR RIAA your using? They seem slightly different to mine, which were gleemed form some circuits I found on the web, but might not be completely accurate.
i started wiht the tango and put it into spice and then tweeked it a bit. I also added the third time constant (50K) below is an excel sheet that lets you scale. the resistor for the third time constant doesn't seem to scale properly and is set for the 6K value.

i'll upload the LT spice .asc if you want it. It is set up so you can step all of the parameters to play.
I'd also be intersted in hearing your choice of valve you had in mind for 6K?
one of the high Gm firecrackers... 6C45, triode 7788, 437... etc.

dave
Attachments
LCR 6k.xls
again a work in progress. R6 is the 50K constant.

The series resistance for L2 was in error and has been corrected.
(97 KiB) Downloaded 1372 times
Last edited by dave slagle on Thu Aug 17, 2006 12:14 pm, edited 2 times in total.
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

I should be able to do some tests on the pentode driver LCR phono I am using, to try and get an idea what the S/N ratio is.
Hi Nick,

Dunno if this helps....

I think Max worked out that the E810F noise was equivalent to a 100R resistor in series with the signal, I assume.

cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

There seems to be some credence given to the possibility that lower impedance RIAA networks might offer better sound,
what is your theory behind this? just curious...
I'd also be intersted in hearing your choice of valve you had in mind for 6K?
one of the high Gm firecrackers... 6C45, triode 7788, 437... etc.

dave
Hi Dave,

Thanks for sharing your work. A high gm firecracker sounds a good bet! If it were me I would drive it as hard as I dare :-) help keep the noise down. James told me that the Russians often designed their valves with this in mind, to keep the noise down, hence they can often be run harder than Western equivalents.

How will you load it?

Just in passing, this may be of use, but I have found that grid leak bias works very well for the second stage and helps keep down Zout. The last two phono's I have built have used this mechanism and I have avoided using a separate line driver.

The lower impedance RIAA idea is something that I don't have a full handle on yet, its more a gut feel. I need to do some more thinking and playing before I can explain it. The idea is not original and I have picked up on it from various sources, but I wouldn't want to incriminate anyone here....I'm sure there's a mathematical proof or denial.

Empirically, I was hoping to do an experiment when I had built a LCR phono and substitute the LCR with an RC of similar quality components, see what the differences are.

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Andrew wrote: How will you load it?
an 80% nickel plate choke of course :-)

Empirically, I was hoping to do an experiment when I had built a LCR phono and substitute the LCR with an RC of similar quality components, see what the differences are.
let me know if you need some help with the chokes.

dave
nickg
Posts: 29
Joined: Sat Mar 19, 2005 1:46 pm

Post by nickg »

I think Max worked out that the E810F noise was equivalent to a 100R resistor in series with the signal, I assume.
Yes, the E810F spec gives its equivilant noise resistance as 110R, however I think that will just be 1/f (it must be, I guess, as its specified by the equiv resistance), I don't think its taking partition noise into account, or any other source.

I am interested in what the entire thing does, after tweeking the supply I have got rid of any visable 50/100hz in the output, so it will be interesting (well to me anyway) to see.

Anyhow, Kaye is away all next week, so it means I can get a new 24 bit sound card through the door without it being noticed (and a pair of 15" drivers for that matter :-))
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

let me know if you need some help with the chokes.

dave
Hi Dave,

Well, that's a kind offer, if I assume that it means you would be willing to wind some, for appropriate payment, of course. What do you envisage the costs to be, roughly speaking?

Question is what to go for? 1k5 looks quite good, has 2uF and 0.05uF caps, the 200R is dead tempting from an experimental point of view and, of course 600R is the standard....decisions decisions....

Time to do some thinking, get some opinions etc..

cheers,

-- Andrew
James D
Posts: 15
Joined: Thu Mar 31, 2005 2:00 pm
Location: UK

Push! Push!

Post by James D »

High Z versus Low Z circuits? Good question.

Things I have noticed - if there is any inductance in the circuit then low Zout sounds better with more micro-dynamics and better tone when impedances are low and, specifically, when driving current is high to match. How low the Zout needs to be and how high the current available for driving from the low Zout needs to be is open to debate...

Funnily enough I have found the same applies to mixed and cap loaded circuits i.e. low Zout and high driving current. Low Zout circuits always seem to be more neutral and have better sense of dynamics.

This seems to apply to the ability to drive the input of the next stage and so needs preserving through any interstage network - so a lower Z RIAA tends to help preserve this ability but at this signal level a low Z stage is probably anything below about 50K! I certainly wouldn't expect to hear any difference between a 200R and a 2K RIAA network all other things being equal (whatever that means)...

It also strikes me as slightly perverse when using custom wound inductors not to choose the inductor values so as to make the cap values preferred values - why make things doubly difficult? If the circuit Zs can be tweaked and the inductors custom values then it surely is sensible to choose these to make the caps easy? There is nothing magical about 600R or 10K or any other network impedance value when you are not constrained by 'standards' that are to apply across many installations and multiple manufacturers...

The 1k5 looks like a nice compromise with the Aikido - it lets you use a 7 times Zout series resistor to swamp any variations in Zout...

Dave, how sure are you that the Tango eq curves are accurate? I guess your not as you tweaked them. What were your tweaks aimed at doing? I guess you calibrated against a reverse RIAA curve in spice to bring it into line? Which reverse RIAA did you use? Just curious!

Nick, We're watching and waiting for the next report!

James
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi James,

Yes 1k5-ish does look good doesn't it. I may take more of a look into this over the next few days.


cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi

Well 1k5 looks good! Here's a snapshot of the LCR network and its freq response when plugged to my phono design, all done in LTSpice. The freq curve doesn't look to bad at all. Inverse RIAA is by Hagerman.

Dave, like James, I'm very interested in your findings/thoughts regarding the LCR values. As you can see, I used your values for L1 and L2, including parasitics but I did tweak some of the other values a bit. Now I'd be the first to say I'm not sure if the changes are better or worse than your posted work in progress, comments, as always, are welcomed.

One thing worth saying is that, it's not as flat as the Tango model I have, but then that doesn't have any parasitics at all and so is a very poor approximation to the real world.

Let me know what you think.

cheers,

-- Andrew
Attachments
Clipboard.jpg
Clipboard.jpg (64.85 KiB) Viewed 44028 times
nickg
Posts: 29
Joined: Sat Mar 19, 2005 1:46 pm

Post by nickg »

OK, I don't know if these are any interest , but I wanted to get some idea of the S/N ratio I am getting out of the E810F Pentode front end in my LCR phono. This is a Denon 103 (0.3mv) into lundahl TX's wired for 1:20. The first graph is the background noise, with the phono stage off. It does show I have some hum to sort out in the wireing to the computer, but as I am more interested in noise, it shouldn't be a problem. All the samples were done in 16bit at 96k.

http://www.lurcher.org/nick/audio/lcr/background.jpg

The next one is playing a 1khz +6db (thats what the test record said) test tone. Hopefully that will give some reference to the noise floor.

http://www.lurcher.org/nick/audio/lcr/+ ... source.jpg

And finally, this is the phono stage with its inputs disconnected, so this should be the noise generated by the phono stage.

http://www.lurcher.org/nick/audio/lcr/no-input.jpg

Hopefully someone who has a clue (not me) can use this to say what the S/N ratio is. But it seems to me to indicate that the pentode isn't doing too much in the way of damage on the noise front.[/u]
Post Reply